Author: Robert Hyatt
Date: 20:16:56 07/07/03
Go up one level in this thread
On July 07, 2003 at 13:35:02, Dieter Buerssner wrote:
>On July 07, 2003 at 10:34:53, Robert Hyatt wrote:
>
>>On July 06, 2003 at 14:05:03, Omid David Tabibi wrote:
>>
>>>On July 06, 2003 at 13:12:25, Robert Hyatt wrote:
>
>>>>#define abs(x) ((x<0) ? -x : x)
>
>>>What's the difference between the above macro and the function
>>>
>>>int simple_abs (int i) {
>>> return (i >= 0) ? i : -i;
>>>}
>>>
>>>when inlined?
>>
>>No difference that I know of, if you have inlining enabled.
>
>Why do you think this? See for example my answer to Omid's question. Some
>problems with the macro can be fixed, by putting the argument x into
>parentheses.
>
>#define abs(x) ((x)<0 ? -(x) : (x))
>
I always do this, as it is a known trick. I didn't type it in my
response however...
The problems with things like abs(++x) I _never_ do. For obvious
reasons. Yes, there are _plenty_ of ways to shoot yourself in the
foot, I don't pass ++ or -- to _any_ function, however.
>As you stated it, for example abs(a+b) will fail (-a+b is in general not
>-(a+b)). The problem with the sideeffects cannot be fixed in Standard C. In many
>cases, slowness, cannot be fixed; for example the compiler may not be able to
>find out, that a function call has no sideeffect.
>
>abs(slow_func_with_no_side_effect(a))
>
>Similar cases, when the argument includes (volatile) global vars.
>
>The function version has no problem. At least with slightly modern compilers, it
>should be best to use the abs() provided by the compiler in most cases. The only
>reason, that comes into mind, that C has this abs() at all, is that it may be
>done fast by tricks (provided by the compiler), that are not available in pure C
>code.
>
>Regards,
>Dieter
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.